Landed cost: the formula, a worked example, and why the number always arrives late

The reason landed cost is hard has nothing to do with the arithmetic. The arithmetic is addition. The difficulty is that the fourteen lines below are billed by eight different parties, arrive over roughly three months, and land in three or four systems that do not talk to each other — and three of them, working capital, FX, and the expedite that gets charged to a premium-freight cost centre instead of the shipment, never reach the SKU at all unless somebody decides to model them.

Two of those lines belong in every model and appear in almost none. The first is expedite: the air freight, the hot-shot truck, the split shipment you bought because a supplier slipped. It models at zero, which is why nobody models it, and when it fires it runs three to eight times the ocean rate on the same volume — charged to a premium-freight cost centre, where it never touches the SKU or the supplier that caused it. If your premium-freight account has a run rate, that run rate is landed cost. The second is freight invoice variance: the accessorials, rebills, and rate-versus-contract gaps that arrive weeks after the move. You are not modelling a rate card, you are modelling what you were actually billed, and on most freight programmes those differ by low single digits of spend before anyone disputes a line.

A few notes on the sizes. Duty is the most variable line, the one people most often guess at, and the one that has moved most since 2024. The MFN rate is only the floor: into the US it runs from zero on a lot of machinery and furniture to over 30% on some apparel and footwear, and Section 301, Section 232 and country-level actions stack on top of the MFN line rather than replacing it — which is how a lane that modelled at 7% ends up entered at 30% with the HTS code unchanged. Two consequences. Get the classification from your broker in writing before you model anything, because a two-digit difference in the tariff line moves landed cost by ten points. And build the model so a rate change is a re-run, not a rebuild. The supplier comparison your team did eighteen months ago is not wrong because the arithmetic was wrong. It is wrong because the duty line under it changed, and nobody re-ran the other forty suppliers.

The last two rows never appear in an ERP landed cost screen, because nobody sends you an invoice for them. On a 30% deposit against an 82-day order-to-receipt cycle at a 9% cost of capital, working capital runs about 1.3% of the FOB value; on a non-functional-currency buy, the move between PO and settlement can be several points more. Both show up in interest expense or in the hedge instead of on the SKU, so neither influences the sourcing decision it should be influencing.

A worked example: the cheaper quote that costs more

You need 150,000 units a year of a stainless steel insulated bottle. Classification, confirmed with your broker, is HTS 9617.00.10, vacuum flasks of one litre or less, at a 7.2% MFN rate. Both suppliers are in Zhejiang, both quote FOB Ningbo, both ship into Los Angeles for drayage to an Inland Empire DC.

Supplier A quotes $6.80 per unit. Supplier B quotes $7.05. On the quote sheet A is 3.5% cheaper, and in most procurement processes the conversation ends there.

Here is the difference the quote sheet does not show. Supplier A ships in retail-ready gift cartons, which fit 4,200 units into a 40' high cube. Supplier B ships in polybags inside 48-count master cartons, which fit 7,500 units into the same box. Both containers cost the same $2,800 to move. Supplier A also runs late on documents, so its container sat three days past free time at the terminal.

For anyone checking the cube: against roughly 76 usable CBM in a 40' high cube, Supplier A's retail gift carton occupies about 18 litres per unit including packaging and Supplier B's polybag-in-master-carton about 10. Neither container weighs out — 7,500 stainless 1L flasks land near 4 tonnes against a payload close to 26 — so this is a pure cubing case, which is the common one. If your product weighs out instead, the same logic holds and the allocation base changes from volume to weight.

Every figure below is per container. Ocean freight here is a mid-cycle assumption. Transpacific spot for a 40' high cube has run below $1,500 in soft markets and into five figures at the 2021-22 peak, and has moved by four figures inside a single month more than once since. Do not model a rate. Model a band, record which rate you used and when you used it, and re-run the comparison when the band moves — because the ranking between two suppliers whose difference is freight-per-unit will flip inside that band. Demurrage is priced at a first-tier $155 per day; tier two and three at LA and Long Beach commonly run $300 to $500 per day, and free time is usually four or five days.

The allocation base changes the answer more than the costs do

Everything above assumed one SKU per container. Real containers are mixed, and the moment they are, you have to decide how to split the freight. Your ERP will offer you a choice: by value, by quantity, by weight, or by volume. That choice is not a formality. It changes per-SKU landed cost by a factor of three.

Take one 40' high cube carrying three SKUs: $45,000 of goods, 68 CBM, 9,000 units, $2,800 of ocean freight to allocate.

SKU-300 is cheap and bulky. Allocate freight by value and it carries $0.137 per unit, which is 6.2% of its $2.20 product cost. Allocate by volume and it carries $0.329, which is 15.0%. Same container, same invoice, and a 2.4× difference in the freight burden on that SKU. SKU-200, which is expensive and dense, moves the other way: $1.369 per unit by value versus $0.412 by volume, a 3.3× swing.

If you are a 3PL or 4PL, this section is your whole landed cost question, phrased differently. Your version is cost-to-serve per client per lane: the same allocation choice decides whether a client's account looks profitable, and whether what you rebill matches what you were billed. Pick the base from how the cost was incurred, hold the container as the join key, and cost-to-serve and rebill accuracy come out of the same record instead of two reconciliations.

Landed cost in NetSuite, SAP, D365, and Odoo

The four ERPs shippers ask about most all handle landed cost, and they handle it in genuinely different ways. Worth knowing which constraint you are working against before you blame the tool. The common thread: all four are allocation engines. Each one will produce a correct landed cost the instant somebody hands it a complete, correctly matched set of costs. Nobody has that set on day 8, and the person who could assemble it is looking at four screens.

Two practical implications. First, whatever ERP you run, decide up front which costs you will estimate at receipt and true up later, and which you will wait for. Estimating freight at a standing rate per container and actualising monthly is almost always better than valuing inventory at PO price and letting the whole 25% arrive as variance. D365's voyage model is built for exactly this; in the others you build it. Second, capture the container number in the ERP at the PO or ASN stage. It is one field, and it is the only key that will later let you tie a forwarder invoice, a broker entry, a demurrage charge, and a warehouse receipt to the same physical box. Teams that skip it spend the next three years reconciling by hand.